The use of automated recidivism risk assessment systems in the u.s. criminal process

This study is devoted to an analysis of the US experience in the use of automated systems of recidivism risk assessment with elements of artificial intelligence in criminal proceedings. It considers the operational principles and scope of application of the algorithms through the example of the most widely used program, COMPAS. It identifies the problems of explanation and non-discrimination nature of decisions made by such programs based on case law and doctrinal discourse and highlights the proposed ways of solving these problems.

As the results of the study, the author's view of the acceptable use of automated systems of recidivism risk assessment, meeting the requirements of due legal process, including in relation to the criminal process of the Russian Federation, is proposed.